Record is a new data type introduced as a preview feature (not certain, but on testing mode) in Java 14, and later it was confirmed in Java 17 as a basic language feature.

A record is a new keyword added in the Java ecosystem to mark it a reserved word, not to be used for any other identifier.

Record is similar to a class but with a few great advantages and a few noticeable disadvantages.

A Record is a datatype and also a class defined in the package java.lang. Whenver you declare an user defined type and specify it as a record, it extends the java.lang.Record internally, hence preventing a record to be extending any other class.

[access-modifier] record <nameOfTheType> (<datatype member1>[, <datatype member2>, ...]) {}

public record Person(String name, int age) {}

We have just created/declared a Person as a record with two attributes - name of type String, and age of type int. That is all. It is now a fullfledged user defined type, ready to use in a normal sense as that of a class, even with the constructors autogenerated, similar to a class.

Mostly, Record is treated as a simplified version of a class that wraps up the much-needed-but-can-be-potentially-avoided boilerplate code in the declaration of the User defined Structure (so far we know it has been defined as a class).

What does it actually wrap it up? In a typical Java class, we have the attributes (members), and behaviors (methods). For each of the attribtutes, we may need the accessors and mutators, or in other words - getters and setters. In addition, we may also need the toString(), hashCode() and equals() - on demand to have a meaningful implementation as desired.

All these put together may certainly elongate the definition of the class, easily about 50 lines of code for 2 members/attributes, making it verbose. A record technically helps the Programmer avoid manually typing all these verbose codes (boilerplate), and rather just focus on the declaration of the attributes, where the rest of all can be automatically taken care.

Techically, for a Java Compiler, we should need all the boiler plate code to work, as we don't have any optimization at that level. However, we just bring these optimization to the Programmer/Developer so that (s)he does not need to get overwhelmed on these trivial and mundante tasks, which otherwise needs time to type OR even get them as an auto-generated code, thanks to the IDE's intelligence. Nevertheless, they visibly become a part of your source code making it too lengthy or verbose.

The record on the other hand generates all the boilerplate code behind the scenes so as to satify the compiler during the compilatin phase for the much needed syntactic evaluvations and parsing. You can compile a record and disassemble the code using javap utility shipped along with the JDK to see all these automagically appearing in the compiled version of the code in the .class file.